I'm wondering if there are any notable instances across the network where the model has spectacularly failed. The criteria I'm curious about are:
- The top voted and accepted answer to the question is incorrect, or at least probably incorrect. (Or, conversely, the correct answer was downvoted into oblivion.)
- The answer was incorrect when posted (i.e., it's wrong and was when they posted it, not just obsolete).
- The question and/or answer has an "adequate" (deliberately fuzzy wording) number of upvotes(/downvotes). This is to prevent instances where the only reason it "failed" was due to obscurity, and also to prevent this from being flooded with 1 upvote incorrect answers to the underwater basketweaving tag.
- The question and answer aren't blatantly off-topic, against the rules, or purely subjective (even "good subjective"). That doesn't mean the answer can't be wrong because of personal bias or agenda, but rather I'm trying to avoid examples where you merely disagree. The "subjective" part may be waived if the answer is of extremely obvious poor quality, but not just because you think they're "wrong".
- It's not just nitpicky wrong. This isn't so we can scrutinize minute details because a 98% correct post made slightly bad assumptions about Java's memory model. It doesn't have to be 100% trash, but use your judgment.
- You should probably log the state of the question and answer when you link it, due to the meta effect possibly (hopefully?) righting these scenarios.
Of course, I didn't say "well known examples", because if an answer was "well known" to be incorrect, it would likely right itself unless it was a concerted troll vote or something. Just something that you, personally, as an expert have noted to be objectively incorrect, but couldn't be corrected merely via your own comment, answer, and/or downvote.
Some may wonder if I have an agenda, but not really. I'm not looking for any examples so I can point and say "ha, the model doesn't work!" I'm just curious, and I think maybe we can learn a bit if we look at the characteristics of cases where it's failed and become better for it.
My running hypothesis is that any instances will be ones that, while they have a verifiable correct answer, nonetheless push certain buttons of a subset of field experts. That or a manifestation of the Fastest Gun in the West Problem where a surreptitiously bad answer stayed on top due to sheer inertia.